Nardo et al. Veterinary Research (2015) 46:77 DOI 10.1186/s13567-015-0213-0 VETERINARY RESEARCH

RESEARCH ARTICLE Open Access Serological profile of foot-and-mouth disease in wildlife populations of West and Central with special reference to Syncerus caffer subspecies Antonello Di Nardo1,2*, Geneviève Libeau3, Bertrand Chardonnet4, Philippe Chardonnet5, Richard A Kock6, Krupali Parekh2, Pip Hamblin2, Yanmin Li2,8, Satya Parida2 and Keith J Sumption7

Abstract The role which West and Central African wildlife populations might play in the transmission dynamics of FMD is not known nor have studies been performed in order to assess the distribution and prevalence of FMD in wild inhabiting those specific regions of Africa. This study reports the FMD serological profile extracted from samples (n =696) collected from wildlife of West and Central Africa between 1999 and 2003. An overall prevalence of FMDV NSP reactive sera of 31.0% (216/696) was estimated, where a significant difference in seropositivity (p = 0.000) was reported for buffalo (64.8%) as opposed to other wild animal species tested (17.8%). Different levels of exposure to the FMDV resulted for each of the buffalo subspecies sampled (p = 0.031): 68.4%, 50.0% and 0% for Nile Buffalo, West and African Forest Buffalo, respectively. The characterisation of the FMDV serotypes tested for buffalo found presence of antibodies against all the six FMDV serotypes tested, although high estimates for type O and SAT 3 were reported for Central Africa. Different patterns of reaction to the six FMDV serotypes tested were recorded, from sera only positive for a single serotype to multiple reactivities. The results confirmed that FMDV circulates in wild populating both West and Central Africa rangelands and in particular in buffalo, also suggesting that multiple FMDV serotypes might be involved with type O, SAT 2 and SAT 1 being dominant. Differences in serotype and spill-over risk between wildlife and livestock likely reflect regional geography, historical circulation and differing trade and livestock systems.

Introduction from a completely inapparent to a rare acutely lethal in- Foot-and-mouth disease (FMD) is an economically dev- fection, making the diagnosis difficult either because the astating disease of intensive livestock farming and high variability in the severity of presenting clinical signs is production , caused by a virus member of the greater than in domestic livestock or because it tends to Apthovirus genus within the Picornaviridae family, and be subclinical for the particular species/virus combin- characterised by an acute and highly contagious vesicu- ation [2,3]. The transmission dynamic of FMD in sub- lar disease which can develop into persistent infection. Saharan Africa is mainly driven by two epidemiological Vesicular lesions resulting from FMD infection are cycles: one in which wildlife plays a significant role in mainly found in tongue, lips and feet but in some cases maintaining and spreading the disease to other suscep- lesions also can occur in snout, muzzle, teats, skin and tible wild and/or domestic ruminants [4-6], whilst with rumen. The disease is characterised by a very short incu- the second the virus is solely transmitted within domes- bation period and high level of virus excretion, particu- tic populations and hence is independent of wildlife. larly in [1]. In wildlife, the FMD pathogenesis varies More than 70 wild animal species have been demon- strated to be susceptible to the FMD virus (FMDV) ei- * Correspondence: [email protected] ther by natural infection or by experimental challenge, 1Institute of Biodiversity, Animal Health and Comparative Medicine, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, and on several occasions the virus has been isolated United Kingdom from naturally infected animals [7]. Among these, Cape 2 The Pirbright Institute, Pirbright, Surrey, Woking, United Kingdom buffalo (Syncerus caffer caffer) has been clearly shown to Full list of author information is available at the end of the article

© 2015 Nardo et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http:// creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Nardo et al. Veterinary Research (2015) 46:77 Page 2 of 16

serve as long-term maintenance host (i.e. carrier) for the seven FMDV serotypes have been reconstructed from Southern African Territories (SAT) FMDV serotypes wildlife samples collected from national parks and faunal [8-10], and in populations of Cape buffalo the virus has reserves of West and Central Africa. The aims of this been estimated to persist for 24 years or longer [11]. In- study were: firstly, to produce an overall picture of the fection in buffalo is subclinical and normally occurs in FMD prevalence in wildlife and mainly in buffalo subspe- calves as soon as maternal antibodies wane at 2–6 cies of West and Central Africa, also characterising risk months of age. Acutely infected buffalo provide sources factors likely associated with the observed prevalence; sec- of infection for other ruminants, both domestic and ondly, to identify the FMDV serotypes potentially circulat- wild, directly or through other species which have ing in resident buffalo populations within the study area. contracted the infection from buffalo [5,6]. Although the In addition, potential limitations of diagnostic testing pro- implication of the buffalo carriers in the epidemiology of cedures used have been evaluated. FMD has not been fully clarified, they have so far been shown to transmit the disease while in that state Materials and methods [4,8,12]. Phylogenetic relationships between SAT types Study population FMDV strains isolated from and those carried by The study was undertaken on serum samples collected from buffalo have been reconstructed from different area of wild ruminants and pigs species during the African Wildlife southern Africa, proving that contacts between livestock Veterinary Project [24], as part of the Pan-African Rinder- and buffalo regularly result in FMD outbreaks among pest Campaign (PARC) and the subsequent programme for cattle [13,14]. Furthermore, available evidence based on the Pan-African Control of Epizootics (PACE) implemented FMDV genome sequencing indicates that (Aepy- in 34 countries across the African continent between 1986 ceros melampus) populations of the Kruger National and 2007. Wildlife species and sampling sites were selected Park – South Africa, usually become infected with SAT at the time according to susceptibility to the Rinderpest viruses derived from buffalo [5]. On occasions SAT line- (RP) virus, population biology (i.e. richness, gregarious ages were demonstrated to have been transmitted first behaviour, and seasonal movements), interface between from buffalo to impala and then from impala to cattle livestock and wildlife, and their availability for veterin- [6,15]. Conceived as such, impala can provide a conduit ary interventions. Sampling was performed using purpos- of infection between buffalo and livestock, acting as an ive sampling by immobilisation, opportunistic sampling by important intermediate between domestic and wild ru- cropping and/or hunting and during field investigations of minants and as an amplifying host in the context of reported episodes of disease and mortality. From the whole FMD transmission dynamics [16]. Presence of antibodies collection stored at the Centre de Coopération Internation- against the FMDV in several wildlife species has been ale en Recherche Agronomique pour le Développement documented in studies conducted in different countries (CIRAD), Montpellier – France, 696 sera collected be- of the African continent, but mainly within its eastern tween 1999 and 2003 were selected as representative of and southern regions [16-18]. In addition, serological wildlife populations present in West and Central Africa screenings implemented in East African countries have (Table 1; Figure 1). In addition, further 19 samples col- indicated potential infections of Cape buffalo with A and lected from cattle within the transfrontier area of the O FMDV serotypes [19-23], although current data do and were included for not support the primary role of buffalo and other wild comparative purpose. Extracted aliquots were sent to The animal species in the transmission of those FMDV sero- Pirbright Institute, Pirbright – United Kingdom (UK), for types generally occurring in domestic ruminants. This diagnostic testing. represents an important pattern of the FMD transmis- sion dynamics in large parts of sub-Saharan Africa that Testing methodology still remains to be explained. Much is already known The sample collection was initially screened for antibodies about the role that Cape buffalo plays in the FMD epi- against the highly conserved NSP of the FMDV using the demiology, largely from studies conducted in South and PrioCHECK® FMDV NS Enzyme-Linked Immunosorbent East Africa; conversely, knowledge on the relationship Assay (ELISA) test kit (Prionics AG, Switzerland), accord- between FMDV and wildlife and/or other buffalo sub- ing to the manufacturer protocol [25]. Specifically, a posi- species that populate the rangelands of western and cen- tive result was considered with a Percentage of Inhibition tral African regions has been less thorough. In order to (PI) value of ≥50, whereas a strong positive result was set progress in the knowledge gap of the FMD epidemiology at a PI value of ≥70. Subsequently, the NS ELISA positive in sub-Saharan Africa and to further investigate the role reactive sera were further assessed using the Solid Phase of wildlife in the transmission of FMD, in this study the Competition ELISA (SPCE) in-house test developed at prevalence of antibodies against the FMDV nonstructural The Pirbright Institute – UK [26,27], thus enabling the protein (NSP) and serological profiles of six out of the qualitative and quantitative characterisation of the specific Nardo et al. Veterinary Research (2015) 46:77 Page 3 of 16

Table 1 Wildlife samples tested allocated by region, country and species of collection. Region Country No. Buffalo Samples No. Other Wildlife TOT West Africa Benin 18 11 29 Burkina Faso 5 30 35 Nigeria 1 7 8 TOT 24 48 72 Central Africa Cameroun - 2 2 Central African Republic 81 247 328 Chad 53 203 256 Democratic Republic of Congo 34 - 34 Gabon 4 - 4 TOT 172 452 624 TOT 196 500 696† †Not including the 19 samples collected from cattle. antibody responses to 6 of the 7 FMDV serotypes (A, O, whilst a value of ≥40 PI was set for SATs serotypes [26]. C, SAT 1, SAT 2 and SAT 3) and, therefore, the FMDV Since the SPCE has not been validated for testing wildlife serotyping profile of each of the serotypes present in sera, the data were further reassessed increasing the cut- Africa at the time of the sampling. The cut-off for the SPCE off to a value of ≥60 PI for serotypes A, O and C, and of was set at a PI value of ≥50 for serotypes A, O and C, ≥50 PI for SATs serotypes to account for an unpredicted

Figure 1 Geographical locations of the wildlife samples selected by species. Distributional extents of buffalo subspecies of the Syncerus genus sourced and adapted from [68]. Nardo et al. Veterinary Research (2015) 46:77 Page 4 of 16

high false positive response. These additional cut-offs were value of 62.0 PI (95%CI 55.1 – 66.8). This figure would re- set as the mean of the single negative response of each flect a defined distinction between the seronegative serotype +5SD as estimated from the original validation (mainly non-buffalo species) and the seropositive (mainly data [26]. According to the above defined thresholds, a buffalo) populations, confirmed by the bimodal distribu- strong positive response was thus considered at either a PI tion found for all species and the left-skewed distribution value of ≥70 or ≥80. Results from the SPCE were con- for the buffalo only. In addition, a total of 39 out of 89 firmed selecting a random sub-sample of the resulting (43.8%) positive samples for the non-buffalo species were SPCE positive reactive sera by Virus Neutralization Test found having a PI value of ≥70 in contrast with the 61.4% (VNT), as prescribed by the World Organisation for Ani- (78/127) estimated for the buffalo population. Thirteen mal Health [28], and using the O1 Manisa, A22 Iraq 24/64, out of 19 samples (68.4%) tested positive for cattle, con- C Phi 7/84, SAT 1 105, SAT 2 Eritrea, SAT 3 309 FMDV firming potential previous exposure of domestic livestock strains. Cut-off for positivity with the VNT was set at a to the FMD; 84.6% (11/13) of those were returning PI titre of ≥1:45, whereas a titre of ≤1:11 and >1:11 but ≤1:32 values of ≥70. were considered as negative and inconclusive, respectively. FMD seroprevalence in wildlife species: descriptive and Data analysis univariate analyses The original database stored at the CIRAD and consisting The NSP testing of the non-buffalo wildlife reported FMD – of information collected through paper forms during the positivity in 89 out of 500 samples (17.8%, 95%CI 14.7% field campaigns was manipulated and inspected for miss- 21.4%), extracted from 16/27 species (59.3%). Among all ing and/or illogical data entries and completed and/or cor- the wildlife species assessed, presence of antibodies were rected whenever possible. The ELISA results were stored found in individuals belonging to the (8.0%), in an Access 2010 (Microsoft Corporation) database along (2.1%), (14.3%), Cephalophinae with associated metadata, such as geographical location (23.5%), Hippotraginae (7.1%), and Reduncinae (27.2%) and GPS coordinates, national park and date of collection, sub-families ( family, 18.4%), and for species be- species and age. Statistical analyses were performed in R longing to the family (11.6%). Bohor 3.1.2 [29], where confidence intervals were calculated (Redunca redunca) (66.8%), Defassa ( using the Agresti-Coull estimation of binomial propor- ellipsiprymnus unctuosus) (63.2%), Red-Flanked tions [30]. Univariate analysis was carried out by the (Cephalophus rufilatus)(60.0%),GiantEland( Adjusted-Wald test, considering the effect of species, age, derbianus) (21.4%) and ( jimela) year of sampling, location and park area on FMD sero- (21.4%) were the wild ruminants reporting high levels of prevalence [31]. All statistically significant variables FMD seropositivity (Table 2). The only sample collected (p < 0.05 two-sided) in the univariate analysis were further from hippopotami and tested positive to the NS ELISA assessed by a generalised linear model (GLM) with a logit (PI = 53) should be regarded as a false positive response to function using a stepwise selection approach, in order to the NS ELISA test, recalling that previous studies con- characterise potential risk factors associated with the ob- ducted in the Kruger National Park failed to detect anti- served FMD seroprevalence. The probability of FMD sero- bodies against the FMDV in this species [2]. prevalence μ was then calculated by back-transforming Effect of the region of collection was found to be egxðÞ of statistical significance on seroprevalence estimates the estimated logit values g(x)asμ ¼ ðÞ [32] and then 1þegx (p = 0.003), where 19% of samples tested positive for Central introduced in a geospatial analysis environment using Arc- Africa whilst presence of antibodies were detected in GIS 10.2.2 (Environmental System Research Institute, only few samples collected in West Africa (3/48). Filter- Inc.) to produce a kernel smoothed intensity map of the ing the results by country of origin, a high FMD preva- predicted FMD prevalence [33]. Pairwise correlation ana- lence was estimated in samples obtained from Chad ’ – lysis based on the Pearsons product moment coefficient (23.1%), although these data were obtained from mostly ρ ( ) was undertaken on all possible combinations of PI esti- a single national park (Zakouma National Park), thus mates resulted from the SPCE testing [34], where missing likely reflecting local conditions. data were treated as pairwise deletions. FMD seroprevalence in buffalo subspecies: descriptive Results and univariate analyses The distributions of PI values resulted from the NS ELISA In total, 127 out of 196 tested positive for FMD (64.8%, test for all the wildlife species (A) and for the buffalo sam- 95%CI 57.9% – 71.1%). According to the subspecies of ples only (B) are plotted in Figure 2. The 50th percentile the Syncerus genus, a high level of NSP antibodies was for all species was reported as 34.2 PI (95%CI 32.0 – 36.3), reported in both Nile Buffalo (Syncerus caffer aequinoc- different from the buffalo distribution that returned a talis) (68.4%) and West African Buffalo (Syncerus caffer Nardo et al. Veterinary Research (2015) 46:77 Page 5 of 16

Figure 2 Histogram and kernel density plots of the NS ELISA percentage of inhibition values estimated for the complete dataset (A) and for buffalo only (B). Red dash-dot line sets the cut-off point (PI = 50).

brachyceros) (50.0%) (Table 3). Although only few sam- The age mean of FMD seropositive individuals was es- ples were tested (0/4), no FMDV reactive sera were timated to be 9.1 ± 5.2 years with high FMD prevalence found for African Forest Buffalo (Syncerus caffer nanus). values described in those animals aged between 2 and The difference in seroprevalence observed between each 10 years. However, no effect (p = 0.542) of age on the of the buffalo sub-species was reported to be significant FMD seropositivity levels extracted from each of the cat- (p = 0.031). egories was observed (Table 3), even though sub-adult Nardo et al. Veterinary Research (2015) 46:77 Page 6 of 16

Table 2 Observed prevalence of FMDV NSP antibodies reported for all the wildlife species tested. Specie No Positive/TOT Observed seroprevalence 95% CI African Bush Elephant (Loxodonta Africana) 0/1 0% - African Forest Buffalo (Syncerus caffer nanus) 0/4 0% - (Philantomba monticola) 1/5 20.0% 2.0% - 64.0% (Redunca redunca) 4/6 66.7% 29.6% - 90.7% ( eurycerus) 0/2 0% - Buffon’s (Kobus kob) 24/172 13.9% 9.5% - 20.0% Bush Duiker (Sylvicapra grimmia) 0/5 0% - Bushbuck (Tragelaphus scriptus) 2/15 13.3% 2.5% - 39.1% Common (Phacochoerus africanus) 4/29 13.8% 4.9% - 31.2% Defassa Waterbuck (Kobus ellipsiprymnus unctuosus) 36/57 63.2% 50.1% - 74.5% Dorcas (Gazella dorcas) 0/40 0% - (Taurotragus derbianus) 3/14 21.4% 6.8% - 48.3% (Hylochoeurs meinertzhageni) 0/1 0% - Greater (Tragelaphus strepsiceros) 0/4 0% - (Alcelaphus buselaphus) 1/20 5.0% 0% - 25.4% (Hippopotamus amphibious) 1/1 100% - Kordofan (Giraffa camelopardis antiquorum) 0/5 0% - Lelwel Hartebeest (Alcelaphus buselaphus lelwel) 2/40 5.0% 0.5% - 17.4% Nile Buffalo (Syncerus caffer aequinoctialis) 115/168 68.4% 61.1% - 75.0% Nolan Warthog (Phacochoerus africanus africanus) 0/7 0% - (Ourebia ourebi) 1/7 14.3% 0.5% - 53.3% ( porcus) 1/6 16.7% 1.1% - 58.2% Red-Flanked Duiker (Cephalophus rufilatus) 3/5 60.0% 22.9% - 88.4% Red-Fronted Gazelle ( rufifrons) 0/1 0% - ( equinus) 2/28 7.1% 0.9% - 23.7% Tiang (Damaliscus korrigum korrigum) 1/5 20.0% 2.0% - 64.0% Topi (Damaliscus korrigum jimela) 3/14 21.4% 6.8% - 48.3% West African Buffalo (Syncerus caffer brachyceros) 12/24 50.0% 31.4% - 68.6% Western Hartebeest (Alcelaphus buselaphus major) 0/8 0% - Yellow-Backed Duiker (Cephalophus silvicultur) 0/2 0% - TOT Buffalo 127/196 64.8% 57.9% - 71.1% TOT Other Wildlife 89/500 17.8% 14.7% - 21.4% TOT Cattle 13/19 68.4% 45.8% - 84.8%

Adjusted-Wald test F(29, 667) = 10.06 (p = 0.000).

and adult categories were those reporting high FMD (80.0%) as opposed to the seroprevalence (83.9%) re- prevalence and narrow interquartile range, which did ported for Central Africa in 2002. not include PI values below the cut-off point. No signifi- No significant difference (p = 0.115) resulted for the cant difference between sexes was observed (p = 0.23). regional prevalence distribution of FMD. Within each Differences in seroprevalence estimates between sam- region, high FMD seroprevalence was found in Burkina pling years were found to be statistically significant Faso (80.0%) for West Africa, and in the Democratic (p = 0.019). Overall, high level of FMD prevalence was Republic of Congo (97.1%) and Central Africa Republic reported in samples collected during 1999 (68.2%) and (64.2%) for Central Africa (Table 3). 2002 (79.0%), although observing a higher seropreva- Considering the area of origin, presence of high levels lence for West Africa in samples collected in 2000 of antibodies against the NSPs was reported in those Nardo et al. Veterinary Research (2015) 46:77 Page 7 of 16

Table 3 Observed prevalence of FMDV NSP antibodies reported for all the buffalo subspecies tested and filtered by age, year, country and park of collection. No Positive/TOT Observed seroprevalence 95% CI Age Group Calf (≤6 m) 1/1 100% - Juvenile (>6 m ≤2ys) 3/5 60.0% 22.9% - 88.4% Sub-adult (>2ys ≤5ys) 36/51 70.6% 56.9% - 81.4% Adult (>5ys) 78/125 62.4% 53.6% - 70.4% Year 1999 30/44 68.2% 53.4% - 80.1% 2000 29/55 52.7% 39.8% - 65.3% 2001 13/22 59.1% 38.7% - 76.8% 2002 49/62 79.0% 67.2% - 87.4% 2003 6/13 46.1% 23.2% - 70.9% Subspecie Nile Buffalo 115/168 68.4% 61.1% - 75.0% West African Buffalo 12/24 50.0% 31.4% - 68.6% African Forest Buffalo 0/4 0% - Country Benin 8/18 44.4% 24.5% - 66.3% Burkina Faso 4/5 80.0% 36.0% - 98.0% Nigeria 0/1 0% - West Africa 12/24 50.0% 31.4% - 68.6% Central African Republic 52/81 64.2% 53.3% - 73.8% Chad 30/53 56.6% 43.3% - 69.1% Democratic Republic of Congo 33/34 97.1% 83.8% - 100% Gabon 0/4 0% - Central Africa 115/172 66.9% 59.5% - 73.5% Park Pendjari National Park 8/18 44.4% 24.5% - 66.3% Pama Reserve 3/3 100% - Arly National Park 1/2 50.0% 9.4% - 90.5% Borgu Game Park 0/1 0% - Manovo-Gounda St. Floris National Park 15/19 78.9% 56.1% - 92.0% Zemongo Faunal Reserve 6/7 85.7% 46.6% - 99.5% Bamingui-Bagoran National Park 3/6 50.0% 18.8% - 81.2% Zakouma National Park 22/41 53.7% 38.7% - 67.9% Garamba National Park 33/34 97.1% 83.8% - 100% Loango National Park 0/4 0% -

Adjusted-Wald test for age group F(3, 178) = 0.72 (p = 0.542). Adjusted-Wald test for year F(4, 192) = 3.03 (p = 0.019). Adjusted-Wald test for subspecie F(2, 194) = 3.52 (p = 0.031). Adjusted-Wald test for country F(6, 190) = 6.62 (p = 0.000). Adjusted-Wald test for region F(1, 195) = 2.51 (p = 0.115). Adjusted-Wald test for park F(9, 126) = 5.27 (p = 0.000). buffalo populations resident in the Garamba National the FMD positive status to the NS ELISA test observed Park (97.1%) of the Democratic Republic of Congo, the for buffalo samples. Increased risk in the probability of Zemongo Reserve (85.7%) and the Manovo-Gounda St. FMD seropositivity was associated with the longitude Floris National Park (78.9%) of the Central African (OR = 1.14, p = 0.011) of the sample locations, whereas a Republic (Table 3). decrease in risk was reported according to the latitude (OR = 0.79, p = 0.000), year (OR = 0.72, p = 0.000) and FMD spatial distribution in buffalo park area (OR = 0.68, p = 0.009) variables entered in the According to the GLM analysis (Table 4), five main ef- model. The kernel smoothed intensity map produced fect variables had statistically detectable association with using the predicted FMD prevalence is shown in Figure 5. Nardo et al. Veterinary Research (2015) 46:77 Page 8 of 16

Table 4 Generalised linear model (logit link) reporting the ORs with corresponding 95% CI for risk factors associated with the FMD seroprevalence reported for all the buffalo subspecies tested. β [95%CI] SE Z p Odds Ratio [95%CI] Intercept 4.11 [1.57 – 6.66] 1.23 3.16 0.002 - Park (Km2) −0.38 [−0.67 – −0.97] 0.15 −2.63 0.009 0.68 [0.51 – 0.91] Longitude 0.13 [0.03 – 0.23] 0.05 2.53 0.011 1.14 [1.03 – 1.26] Latitude −0.23 [−0.37 – −0.10] 0.07 −3.50 0.000 0.79 [0.69 – 0.9] Year −0.32 [−0.48 – −0.17] 0.08 −4.10 0.000 0.72 [0.62 – 0.84] Age 0.01 [0 – 0.01] 0.002 4.31 0.000 1.01 [1 – 1.01] log-likelihood = −479.58; AIC = 973.17.

The areas with high risk of FMD amongst the buffalo pop- Africa samples was set below the threshold values and the ulations sampled, as predicted by the model, are located data distributions of type A and O were largely right- mainly in the bordering areas between south-west Chad skewed (with most of the data lying below the cut-off and north-west Central African Republic, and in the points), in contrast with what reported for Central Africa. north-east Democratic Republic of Congo that borders Different patterns of reaction to the 6 FMDV serotypes with South . This spatial range overlaps with the ex- tested were recorded, from sera only positive for a single tent of the Aouk and Zakouma National Parks and the serotype to multiple reactivities. A number of sera with Aouk Aoukale Faunal Reserve in Chad, the Manovo- the highest serotype-specific responses (i.e. highest PI Gounda St. Floris and Bamingui-Bangoran National Parks values) were identified for type O (16.7%, 2/12), C in the Central African Republic. The Manouvo-Gounda (16.7%, 2/12) and SAT 2 (58.3%, 7/12) in samples col- St. Floris and the Zakouma National Parks constitute the lected from West Africa, and for type A (2.6%, 3/115), O same ecological area and, as evidenced by the model pre- (47.8%, 55/115), C (5.2%, 6/115), SAT 1 (14.8%, 17/115), diction and the prevalence reported, it is likely to be SAT 2 (27.0%, 31/115) in those retrieved from Central regarded as high risk of FMD. Africa (Table 6). No sera with the highest serotype-specific response for SAT 3 were reported, even though PI values of up to 79 and 87 were estimated from samples of West FMDV serotyping profile in buffalo: descriptive and and Central Africa, respectively. The potential cross- correlation analyses reaction between pairs of serotypes tested was then The overall results from the SPCE analysis of buffalo assessed computing the pairwise correlation matrix of samples showed presence of antibodies against all the 6 continuous data (PI values) for all the samples analysed FMDV serotypes tested (Figure 3), with high levels esti- (Figure 4). Statistically significant correlations (p =0.000) mated for O (82.3%), SAT 2 (81.9%) and SAT 1 (73.2%) with high ρ coefficients were reported for the A–SAT 1 serotypes (Table 5). Increasing the cut-off as described (0.66), A–SAT 3 (0.61), SAT 1–SAT 3 (0.70), C–SAT 3 in the methodology section did not largely change the (0.67) and C–SAT 2 (0.54) pairs. No correlation was found seroprevalence patterns found for O, SAT 1 and SAT 2 between O and any of the other FMDV serotypes tested FMDV serotypes, differently from the A, C and SAT 3 (ρ≤ 0.1; p > 0.05). The random sample (n = 43) extracted seropositivities which were reduced to the order of 50%. from the SPCE positive data was confirmed by the VNT In addition, 41 out of 102 positive samples (40.2%), 44/ test, which reported positive results at the highest titre of 104 (42.3%) and 19/93 (20.4%) returned PI values ≥80 1:90, 1:178, 1:256, 1:1024 and 1:355 for O, C, SAT 1, SAT for type O, SAT 2 and SAT 1, respectively. 2 and SAT 3 FMDV serotypes, respectively. Inconclusive Interestingly, the pattern of FMD prevalence for each results were obtained for type A (titre of 1:22). of the 6 serotypes tested was shaped differently accord- ing to the region of collection. Although high level of antibody responses against both SAT 1 and SAT 2 Discussion FMDV serotypes were recorded for both West and Cen- This study reports the FMD serological profile extracted tral Africa, higher prevalence of type O, C and SAT 3 from wildlife populations inhabiting the rangelands of were found in samples collected from Central African West and Central Africa. The results confirm that countries. Moreover, considering the distribution of PI FMDV circulates within wildlife-livestock ecosystems values returned for each of the 6 serotypes by region present in the study regions and in particular in buffalo (Figure 3), the third quartile (Q3) of type A, O, SAT 3 subspecies, also suggesting that multiple FMDV sero- and to some extent of C results obtained for the West types may be involved with type O, SAT 2 and SAT 1 Nardo et al. Veterinary Research (2015) 46:77 Page 9 of 16

Figure 3 Box plot of the SPCE percentage of inhibition values for buffalo samples according to each of the FMDV serotypes tested [Overall (A) and regional (B) data]. Cut-off was set at either a PI value of ≥50 or ≥60 for A, O and C serotypes, and at either a PI value of ≥40 or ≥50 for SATs serotypes (red dotted lines) outlier. being dominant. A different pattern of FMD prevalence FMDV serotypes circulating in buffalo populations for each of the serotypes tested was reported between present in West and Central Africa, which might be as- West and Central Africa, with high levels of serotype- sociated with transboundary movements of FMDV line- specific antibodies against type O, C and SAT 3 found in ages and, thus, in line with what has been historically buffalo samples sourced from Central Africa. These re- described for the FMDV pools 4 and 5 [35,36]. Although sults would indicate a distinct geographical extent of historical data of FMDV isolates recovered from buffalo Nardo et al. Veterinary Research (2015) 46:77 Page 10 of 16

Figure 4 Scatterplot matrix of the pairwise correlation analysis estimated between PI values obtained from buffalo samples tested for each of the FMDV serotypes by SPCE. Variables are reordered and coloured according to the returned Pearson correlation values [blue (ρ ≤ 0.3); yellow (0.3 >ρ ≤ 0.5); red (ρ ≥ 0.5)], where higher correlated variables are plot near the diagonal. samples within the countries under study at the time of 2005 and 2012–13), Mali (O in 2004–05; A in 2004 and the sampling were generally not available, the here de- 2006), Mauritania (O in 2000–01; A in 2006), Niger (O fined FMD distribution in wild ruminants of West and in 2001 and 2005; SAT 2 in 2007–08 and 2011–12), Central Africa might largely contribute in providing a Senegal (O in 2001 and 2006; SAT 2 in 2009), Togo (O clearer picture of the FMD burden in those largely un- in 2004–05; A in 2005) [37-43]. From 1970, reports of studied regions of sub-Saharan Africa. The reported out- FMD activities in Central Africa were only available for breaks affecting livestock of West Africa since 2000 types A and SAT 2 in Chad (1973 and 1972, respect- were caused by FMDV types A, O and SAT 2 circulating ively) and for the Democratic Republic of Congo (O in in Benin (O and A in 2010), Burkina Faso (O in 2002), 2006 and 2010; A in 2011; SAT 2 in 1974, 1979 and Cameroon (O in 2000 and 2005; A and SAT 2 in 2000, 1982) [44]. FMDV type C has never been reported in Nardo et al. Veterinary Research (2015) 46:77 Page 11 of 16

Table 5 Observed prevalence of serotype-specific FMDV antibodies reported for all the buffalo subspecies tested as overall result and by region of collection. Serotype No Positive/TOT Observed seroprevalence† 95%CI West Africa A 6/12 50.0% 25.4% - 74.6% O 6/12 50.0% 25.4% - 74.6% C 7/12 58.3% 31.9% - 80.7% SAT 1 8/12 66.7% 38.8% - 86.4% SAT 2 9/12 75.0% 46.1% - 91.7% SAT 3 3/12 25.0% 8.3% - 53.8% Central Africa A 50/115 43.4% 34.8% - 52.6% O 96/112 85.7% 77.9% - 91.1% C 77/115 67.0% 57.9% - 74.9% SAT 1 85/115 73.9% 65.2% - 81.1% SAT 2 95/115 82.6% 74.6% - 88.5% SAT 3 56/115 48.7% 39.7% - 57.7% TOT A 56/127 44.1% 35.8% - 52.8% O 102/124 82.3% 74.5% - 88.0% C 84/127 66.1% 57.5% - 73.8% SAT 1 93/127 73.2% 64.9% - 80.2% SAT 2 104/127 81.9% 74.2% - 87.7% SAT 3 59/127 46.5% 38.0% - 55.1% †Cut-off values set as ≥50 for A, O and C serotypes, and ≥40 for SATs serotypes.

West and Central Africa and, moreover, up until 2004 part of the Queen Elizabeth National Park [46,47]. In 2013, this FMDV serotype was only confined to East Africa in FMDV SAT 3 was isolated from a sub-clinically (or persist- (1957–2004), (1957–1983) and ently) infected Ankole calf at Nyakatonzi (Kasese District), (1970–71), from when it seems to have been extinct in close proximity to the northern part of the Queen Eliza- [45]. A very striking results provided from this study is beth National Park [48]. These finding may indicate that the relatively high proportion (84/127) of serotype C this serotype is also potentially maintained in buffalo popu- positive sera, although with only 6 producing very high lations present in wildlife ecosystems of Eastern Africa [17]. PI values. This may either indicate a potential cross- The SPCE test used in this study for the qualitative reactivity with other serotypes (high correlations were and quantitative detection of antibodies against the found between C and both SAT 2 and SAT 3) or that FMDV serotypes has previously proven to be more ro- serotype C may have been circulating without being de- bust and specific, and equally sensitive to the Liquid tected, with the latter less plausible. The SAT 3 serotype Phase Blocking ELISA (LPBE) [26], but not totally un- has been mainly documented in Southern African coun- affected by serological cross-reactivity between FMDV tries with occasional isolations in Uganda (1970 and 1997) serotypes, thus representing a valid, easy and fast to from samples collected from Cape Buffalo in the southern process alternative to the VNT. It should be noted that

Table 6 Number of buffalo sera with highest serotype-specific FMDV antibodies response [highest PI value] per serotype tested positive on the SPCE by country of collection. A O C SAT 1 SAT 2 SAT 3 Benin 0/8 [60] 2/8 [73] 2/8 [92] 0/8 [87] 3/8 [92] 0/8 [79] Burkina Faso 0/4 [63] 0/4 [67] 0/4 [71] 0/4 [76] 4/4 [99] 0/4 [40] West Africa 0/12 [63] 2/12 [73] 2/12 [92] 0/12 [87] 7/12 [99] 0/12 [79] Central African Republic 1/52 [80] 22/52 [95] 4/52 [85] 11/52 [98] 13/52 [93] 0/52 [87] Chad 2/30 [75] 9/30 [85] 2/30 [74] 2/30 [87] 13/30 [96] 0/30 [81] Democratic Republic of Congo 0/33 [82] 24/33 [108] 0/33 [88] 4/33 [97] 5/33 [90] 0/33 [83] Central Africa 3/115 [82] 55/115 [108] 6/115 [88] 17/115 [98] 31/115 [96] 0/115 [87] Nardo et al. Veterinary Research (2015) 46:77 Page 12 of 16

Figure 5 Kernel density map of the predicted probability μ of FMD seropositivity in buffalo as estimated from the generalised linear model (logit link). Geographical extent of African wildlife protected areas are sourced and adapted from [69]. the SPCE has not been validated for testing wildlife sera, ruled out. However, subsequent multiple serotypes infec- but only for cattle, pigs and sheep [26,27], and this might tions do occur and even simultaneous multiple infection have an impact on the results here generated. However, cannot be excluded. In a previous study it has been dem- the selection of higher cut-off values for the SPCE positiv- onstrated that buffalo carriers are refractory to reinfection ity threshold, aiming at reducing a potential false positive with the same strain of virus [49,50] thus supporting the response, did not change the seroprevalence figures here hypothesis of potential co-infections and/or subsequent reported and, furthermore, several buffalo samples infections with more than one serotypes. Although the re- returned PI values of ≥80 thus indicating presence of high sults here reported suggest infections of buffalo subspecies antibody levels. In addition, a random sample of the posi- with different FMDV serotypes, there is little published on tive sera generated from the SPCE has been confirmed by non-SAT serotypes in buffalo population [19-23], with no VNT testing. The mismatch between the strains used for evidences supporting the hypothesis that domestic types the VNT testing and antibodies present in the sera might of FMDV has come from buffalo as carrier. In a recent have an impact on the results obtained. However, the pre- study, the phylogenetic descent of the SAT 2 serotype cise selection of an appropriate test antigen would not be across the entire African continent was regarded to have possible when testing sera of unknown status and, there- originated from a FMDV ancestor formerly infecting Cape fore, it has been assumed that antigenic differences were buffalo, also evaluating that interspecies virus transitions limited, with positive samples still producing high titres to might occur between Cape buffalo and cattle, and vice the specific strains than to the other serotypes. Although versa [51]. Although this would support the hypothesis correlations between serotype-specific responses have of FMDV shifting between wildlife and livestock, the been evaluated, the extent to which findings of individual incomplete and biased nature of the analysed data sera reactive to multiple serotypes is due to serological could potentially derive confounding results. Neverthe- cross-reactivity or multiple infections has not been entirely less, genetic and epidemiological analyses have clearly Nardo et al. Veterinary Research (2015) 46:77 Page 13 of 16

shown on specific occasions the close relationship be- habitat of Cape buffalo in Eastern and Southern African tween SATs viruses infecting buffalo and other wild regions. Although the predicted spatial distribution of species (e.g. Impala) to those causing outbreaks in cat- FMD in buffalo might reflect the epidemiological status at tle [16,52,53]. In addition, a field study conducted in the time of the sampling, a generalisation to the current Ethiopia found significant association between cattle situation might be expected since no effective control exposed to FMDV and their contact history with wild- measures have been implemented in recent years either in life [54], whilst a recent study conducted at the periph- West or Central African regions, and up-to-date data have ery of protected areas in Zimbabwe indicates that not been published. In addition, although accounting for interactions between livestock and buffalo populations the extent of the protected areas present in West and can account for FMD primary outbreaks [55]. Besides Central Africa, the predicted FMD spatial distribution was studies based on serological investigations, to what ex- not intended to resolve the landscape structure of the tent types of FMDV prevalent in domestic ruminants study regions, and so this might cover areas where buffalo infect wildlife is unknown; hence, until this issue is in- ecological niches might be absent. Therefore, a more vestigated thoroughly it will constitute a major defi- extensive sampling frame would be required to provide a ciency in understanding the epidemiology of FMD in more exhaustive indication of the spatial burden of FMD in large parts of sub-Saharan Africa. Therefore, further wildlife ecosystems of West and Central Africa. This initial field studies are warranted to collect clinical samples, in attempt would, nevertheless, provide useful information order to enable the genome characterisation of FMDV to be linked for a broad scale strategic FMD monitoring lineages circulating within wild animal species of West planning. and Central Africa and to confirm if FMDV serotypes nor- According to the buffalo subspecies tested, the FMD mally present in domestic livestock are really mixing with prevalence found might reflect the different social buffalo and, thus, eventually become established in those organization and population biology of each of the sub- populations. However, it should be noticed that two species considered. For example, the African forest buf- significant buffalo populations at present exist in West falo herds are quite isolated, have a small animal density and Central Africa, which are susceptible to share cattle (herd size of about 3–25 individuals) and move in a lim- grounds (Figures 1 and 5): the W-Arly-Pendjari (WAP) ited home-range (~2.3–8km2) compared to the Cape Parks Complex (tranfrontier area shared between Benin, buffalo. In addition, they have a very secretive behaviour Burkina Faso and Niger, which hosts around 10000+ buf- and a limited distribution, mainly found in rainforest falo) and the Zakouma National Park (which hosts around ecosystem [57]. Savannah buffalo (West African and Nile 7000+ buffalo). This geographic distribution of buffalo buffalo) more frequently split into smaller herds, herd- would therefore likely contribute to reduce the mixing be- switching is more common and they have a large home tween buffalo and livestock thus decreasing the risk of range. This is opposed to Cape buffalo herds that are FMD transmission. more densely populated (e.g. mass herds can vary from Wild animal species already reported to be susceptible few hundreds up to several thousand individuals) and to the FMDV [7] were confirmed in this study. In engage in long-distance dispersal [58]. In a previous addition, presence of antibodies against the FMDV has study conducted in Cape buffalo population of East not been previously described in Buffon’s Kob (Kobus Africa, a 67.7% of the total samples tested was reported kob) and Oribi (Ourebia ourebi), therefore increasing the as FMD positive [17]. Therefore, this might indicate dif- number of wild species reported to be suscep- ferent patterns of FMD susceptibility across the different tible to FMD infection in sub-Saharan Africa. In addition, buffalo subspecies that might be linked to the species the FMD prevalence found in non-buffalo species of West ecology, even though only four African forest buffalo and Central Africa is higher than what has previously been samples were available for testing. In fact, susceptibilities reported in other studies targeting wildlife ecosystems of to the FMDV of buffalo living in savannah habitat are Eastern Africa and Zimbabwe [17,18,56]. The high FMD consistent across sub-species but African forest buffalo prevalence found in Bohor Reedbuck and Waterbuck might show a different epidemiology. On the other hand, might reflect their ecology and living ecosystem: in fact this could be a concurrent result of the host living habitat, they are dependent on water with living habitats close to the social behaviour of each of the buffalo subspecies, and water sources, which might indicate a link between FMD the extent of the FMD geographical distribution (i.e. the transmission within and between wildlife species (and/or African forest buffalo samples were collected from Gabon, between domestic and wild animals) when congregating at in which FMD has never been reported in the periods watering points. This study provides the first evidence of between 1996 and 2003 [59], and between 2006 and 2012 FMDV exposure in subspecies of the Syncerus genus other [60]). In addition, it should be noted that the African for- than the Cape buffalo [7], thus accounting for their poten- est buffalo were darted in the Loango National Park where tial role in the epidemiology of FMD outside the living the livestock presence is very low and especially absent Nardo et al. Veterinary Research (2015) 46:77 Page 14 of 16

from the darting area, that means lack of contacts between The control of the wildlife-livestock interface in the buffalo and cattle besides the very low size of herds (<8 transmission of FMD has been only successfully applied in heads). From the model estimates, the odds of a buffalo South Africa at a considerable ecological and economic resulting FMD positive by NS ELISA is reduced by 0.98 cost [64,65] by the means of herds separation (e.g. strict for every 10 km2 increase in the park area. This would be land-use policies, animal movement controls and fencing) directly correlated with the density of buffalo population, and buffer vaccination of livestock population around the which might indicate that in large ecosystems herds tend source of virus [66]. However, in countries where wildlife to be sparser, not overlapping their home ranges, and thus populations are integrated with extensive nomadic and diminishing the risk of transmitting and maintaining the semi-nomadic pastoralism, as would be the case for West disease. In addition, the rainforest area of western and and Central Africa, the risk of FMD spread increases not central Africa represents a limiting factor for the spatial only for the difficulty in applying effective control mea- and demographic expansion of buffalo herds (and live- sures (e.g. pastoralists usually rely on ethnoveterinary stock), as would be the case for African forest buffalo that practices [67] and move within and between countries in mainly inhabits forest clearings [61]. Figures indicate that an uncontrolled fashion), but also for the increase in the 89.3% of rainforests is present in Central Africa, where the land-use pressure and conflict between pastoralists and Democratic Republic of Congo accounts for the largest wildlife competing for grazing and water resources. As African lowland rainforest area (53.6%) [62]. However, the example, the St. Floris National Park, Central African FMD prevalence (97.1%) obtained from the Democratic Republic, is boasting a large population of buffalo and is Republic of Congo was resulting from Nile buffalo sam- a grazing and transhumance crossing land for the Fulani ples collected in the Garamba National Park, which pastoralist tribe. A 78.9% of FMD prevalence in Nile buf- mainly covers vast grass savannahs and woodlands. Fur- falo has been reported in this ecosystem, with high thermore, density of buffalo population fluctuates accord- serotype-positive responses against type O, SAT 1 and ing to seasons, which tend to be reduced during the dry SAT 2. The demographic growth, the expansion of cultiva- season as at this time of the year the size of the area for tion (e.g. agro-pastoral systems), the development of local grazing and watering is really reduced compared to the governance on natural resources [65] and the reduction of rainy season. rangeland resources in Africa have indeed led to increased Direct contacts between buffalo and livestock seem sharing of resources between domestic and wildlife species to be rare, with degrees of variability determined by and hence the risk of diseases transmission. However, the ecosystem structure and climatic cycles. It seems to be ecological processes driving the FMDV evolution and more common in open habitats and with plains species transmission in the sub-Saharan African ecosystems still (e.g. during mass migration of , topi, zebra remain poorly documented. In this context the nature of and as in the Mara-Serengeti ecosystem in East the wildlife-livestock interface in pastoral landscapes and Africa) and during the dry seasons, when water avail- its complex socio-ecological-economic interactions have ability constrains animals to closely congregate thus in- not been entirely studied and need to be thoroughly inves- creasing the transmission likelihood of water-related tigated. The importance and, at the same time, the diffi- infections [63]; in fact, these are also the times of the culties in controlling the FMD in the sub-Saharan Africa highest level of residency and density within protected relies on the unique diversity and numbers of the wild spe- areas of both livestock and buffalo. In addition, it cies present and the explosive growth of the human popu- should be noted that wildlife usually avoids livestock lation, which consequently needs to create solutions that and human contacts (i.e. at watering points or locations would work for improving the agriculture standards, the with key forage resources) in a space-time fashion un- sanitary safety of livestock trade, the sustainable land use less habituated. As FMD requires a relatively close con- and the biodiversity conservation of wildlife ecosystems. tact setting for interspecies transmission, the FMD Competing interests interface between wildlife and livestock should, there- The authors declare that they have no competing interests. fore, not be seen as a direct physical interaction but as an indirect contact (i.e. through soil, forage and water Authors’ contributions ADN and KJS conceived and designed the study. BC, PC and RAK implemented contaminated by bodily discharge of infected animals), the original field surveys and collected the samples. ADN, KP and PH processed which might be regarded as the most likely factor to be and tested the samples. GL, SP and YL participated in the coordination of the associated with the risk of FMDV transfer from domes- diagnostic testing. ADN performed the statistical analysis. ADN drafted and BC, PC, RAK and PH reviewed the manuscript. All authors read and approved the tic to wildlife species. Eventually, the spread of FMD final manuscript. within the wildlife-domestic interface might be driven by a complex interplay of risk factors, including bio- Acknowledgements Thanks to Nick J Knowles for providing data on historical FMDV outbreaks. physical and climatic features, ecological traits and hu- This study was supported by the European Commission for the Control of man practices. Foot-and-Mouth Disease (EuFMD) under the Food and Agriculture Nardo et al. Veterinary Research (2015) 46:77 Page 15 of 16

Organization of the United Nations (FAO) umbrella. Laboratory testing at The 15. Hargreaves SK, Foggin CM, Anderson EC, Bastos AD, Thomson GR, Ferris NP, Pirbright Institute were performed under the auspices of the World Reference Knowles NJ (2004) An investigation into the source and spread of foot and Laboratory for Foot-and-mouth disease (WRLFMD), supported with funding mouth disease virus from a wildlife conservancy in Zimbabwe. Rev Sci Tech from the European Union. The views expressed herein can in no way be taken 23:783–790 to reflect the official opinion of the European Union. 16. Vosloo W, Thompson PN, Botha B, Bengis RG, Thomson GR (2009) Longitudinal study to investigate the role of impala (Aepyceros melampus) Author details in foot-and-mouth disease maintenance in the Kruger National Park, South 1Institute of Biodiversity, Animal Health and Comparative Medicine, College Africa. Transbound Emerg Dis 56:18–30 of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, 17. Bronsvoort BM, Parida S, Handel I, McFarland S, Fleming L, Hamblin P, Kock R United Kingdom. 2The Pirbright Institute, Pirbright, Surrey, Woking, United (2008) Serological survey for foot-and-mouth disease virus in wildlife in eastern Kingdom. 3Centre de Coopération Internationale en Recherche Africa and estimation of test parameters of a nonstructural protein enzyme- Agronomique pour le Développement (CIRAD), UMR Contrôle des Maladies, linked immunosorbent assay for buffalo. Clin Vaccine Immunol 15:1003–1011 Campus International de Baillarguet, Montpellier, France. 4African Protected 18. Condy JB, Herniman KA, Hedger RS (1969) Foot-and-mouth disease in Areas & Wildlife Expert, Saint Cloud, France. 5International Foundation for the wildlife in Rhodesia and other African territories. A serological survey. Conservation of Wildlife (IGF), Paris, France. 6Royal Veterinary College (RVC), J Comp Pathol 79:27–31 Hatfield, United Kingdom. 7European Commission for the Control of 19. Ayebazibwe C, Mwiine FN, Balinda SN, Tjornehoj K, Masembe C, Muwanika VB, Foot-and-Mouth Disease (EuFMD), Food and Agriculture Organisation of the Okurut AR, Siegismund HR, Alexandersen S (2010) Antibodies against foot-and- United Nations (FAO), Rome, Italy. 8Present address: Boehringer Ingelheim mouth disease (FMD) virus in African buffalos (Syncerus caffer) in selected GmbH, Shanghai, China. National Parks in Uganda (2001–2003). Transbound Emerg Dis 57:286–292 20. Ayebazibwe C, Mwiine FN, Tjornehoj K, Balinda SN, Muwanika VB, Ademun Received: 21 January 2015 Accepted: 9 June 2015 Okurut AR, Belsham GJ, Normann P, Siegismund HR, Alexandersen S (2010) The role of African buffalos (Syncerus caffer) in the maintenance of foot- and-mouth disease in Uganda. BMC Vet Res 6:54 21. Kalema-Zikusoka G, Bengis R, Michel A, Woodford M (2005) A preliminary References investigation of tuberculosis and other diseases in African buffalo (Syncerus 1. Alexandersen S, Zhang Z, Donaldson AI, Garland AJ (2003) The caffer) in Queen Elizabeth National Park, Uganda. Onderstepoort J Vet Res pathogenesis and diagnosis of foot-and-mouth disease. J Comp Pathol 72:145–151 129:1–36 22. Hamblin C, Anderson EC, Jago M, Mlengeya T, Hipji K (1990) Antibodies to 2. Thomson GR, Vosloo W, Bastos AD (2003) Foot and mouth disease in some pathogenic agents in free-living wild species in . Epidemiol wildlife. Virus Res 91:145–161 Infect 105:585–594 3. Arzt J, Baxt B, Grubman MJ, Jackson T, Juleff N, Rhyan J, Rieder E, Waters R, Rodriguez LL (2011) The pathogenesis of foot-and-mouth disease II: viral 23. Anderson EC, Doughty WJ, Anderson J, Paling R (1979) The pathogenesis of pathways in swine, small ruminants, and wildlife; myotropism, chronic foot-and-mouth disease in the African buffalo (Syncerus caffer) and the role syndromes, and molecular virus-host interactions. Transbound Emerg Dis of this species in the epidemiology of the disease in Kenya. J Comp Pathol – 58:305–326 89:541 549 4. Dawe PS, Flanagan FO, Madekurozwa RL, Sorensen KJ, Anderson EC, Foggin 24. Chardonnet P, Kock R (2001) Final report of the African Wildlife Veterinary CM, Ferris NP, Knowles NJ (1994) Natural transmission of foot-and-mouth Project (November 1998 - June 2000). CIRAD-EMVT, Montpellier, France disease virus from African buffalo (Syncerus caffer) to cattle in a wildlife area 25. Sorensen KJ, de Stricker K, Dyrting KC, Grazioli S, Haas B (2005) of Zimbabwe. Vet Rec 134:230–232 Differentiation of foot-and-mouth disease virus infected animals from 5. Bastos A, Boshoff CI, Keet DF, Bangis RG, Thomson GR (2000) Natural vaccinated animals using a blocking ELISA based on baculovirus transmission of foot-and-mouth disease virus between African buffalo expressed FMDV 3ABC antigen and a 3ABC monoclonal antibody. Arch – (Syncerus caffer) and impala (Aepyceros melampus) in the Kruger National Virol 150:805 814 Park, South Africa. Epidemiol Infect 124:591 26. Li Y, Swabey KG, Gibson D, Keel PJ, Hamblin P, Wilsden G, Corteyn M, Ferris 6. Sutmoller P, Thomson GR, Hargreaves SK, Foggin CM, Anderson EC (2000) NP (2012) Evaluation of the solid phase competition ELISA for detecting The foot and mouth disease risk posed by African buffalo within wildlife antibodies against the six foot-and-mouth disease virus non-O serotypes. – conservancies to the cattle industry of Zimbabwe. Prev Vet Med 44:43–60 J Virol Methods 183:125 131 7. Weaver GV, Domenech J, Thierman AR, Karesh WB (2013) Foot and mouth 27. Paiba GA, Anderson J, Paton DJ, Soldan AW, Alexandersen S, Corteyn M, disease: a look from the wild side. J Wild Dis 49:759–785 Wilsden G, Hamblin P, MacKay DK, Donaldson AI (2004) Validation of a foot- 8. Vosloo W, Bastos AD, Kirkbride E, Esterhuysen JJ, van Rensburg DJ, Bengis and-mouth disease antibody screening solid-phase competition ELISA – RG, Keet DW, Thomson GR (1996) Persistent infection of African buffalo (SPCE). J Virol Methods 115:145 158 (Syncerus caffer) with SAT-type foot-and-mouth disease viruses: rate of 28. Golding SM, Hedger RS, Talbot P (1976) Radial immuno-diffusion and fixation of mutations, antigenic change and interspecies transmission. J Gen serum-neutralisation techniques for the assay of antibodies to swine vesicular – Virol 77:1457–1467 disease. Res Vet Sci 20:142 147 9. Thomson GR, Vosloo W, Esterhuysen JJ, Bengis RG (1992) Maintenance of 29. Core Team R (2015) R, A language and environment for statistical foot and mouth disease viruses in buffalo (Syncerus caffer Sparrman, 1779) computing. R Foundation for Statistical Computing, Vienna, Austria in southern Africa. Rev Sci Tech 11:1097–1107 30. Brown L, Cai T, DasGupta A (2001) Interval estimation for a binomial 10. Hedger RS (1972) Foot-and-mouth disease and the African buffalo (Syncerus proportion. Stat Sci 16:101–133 caffer). J Comp Pathol 82:19–28 31. Koch G, Freeman GDH, Freeman JL (1975) Strategies in the multivariate 11. Condy JB, Hedger RS, Hamblin C, Barnett IT (1985) The duration of the foot- analysis of data from complex surveys. Int Stat Rev 43:59–78 and-mouth disease virus carrier state in African buffalo (i) in the individual 32. Hosmer D, Lemeshow S (2000) Applied logistic regression, 2nd edn. John animal and (ii) in a free-living herd. Comp Immunol Microbiol Infect Dis Wiley and Sons Ltd., New York 8:259–265 33. Silverman B (1986) Density estimation for statistics and data analysis. 12. Dawe PS, Sorensen K, Ferris NP, Barnett IT, Armstrong RM, Knowles NJ Chapman and Hall, New York (1994) Experimental transmission of foot-and-mouth disease virus from 34. Kirkwood B, Sterne J (2003) Essential medical statistics, 2nd edn. Blackwell carrier African buffalo (Syncerus caffer) to cattle in Zimbabwe. Vet Rec Publishing Ltd., Oxford, UK 134:211–215 35. Di Nardo A, Knowles NJ, Paton DJ (2011) Combining livestock trade 13. Bruckner GK, Vosloo W, Du Plessis BJ, Kloeck PE, Connoway L, Ekron MD, patterns with phylogenetics to help understand the spread of foot and Weaver DB, Dickason CJ, Schreuder FJ, Marais T, Mogajane ME (2002) Foot mouth disease in sub-Saharan Africa, the Middle East and Southeast Asia. and mouth disease: the experience of South Africa. Rev Sci Tech 21:751–764 Rev Sci Tech 30:63–85 14. Vosloo W, Boshoff K, Dwarka R, Bastos A (2002) The possible role that 36. Rweyemamu M, Roeder P, Mackay D, Sumption K, Brownlie J, Leforban Y, buffalo played in the recent outbreaks of foot-and-mouth disease in South Valarcher JF, Knowles NJ, Saraiva V (2008) Epidemiological patterns of foot- Africa. Ann NY Acad Sci 969:187–190 and-mouth disease worldwide. Transbound Emerg Dis 55:57–72 Nardo et al. Veterinary Research (2015) 46:77 Page 16 of 16

37. Sangare O, Bastos AD, Venter EH, Vosloo W (2003) Retrospective genetic 57. Melletti M, Penteriani V, Mirabile M, Boitani L (2007) Some behavioral analysis of SAT-1 type foot-and-mouth disease outbreaks in West Africa aspects of forest buffalo (syncerus caffer nanus): from herd to individual. (1975–1981). Vet Microbiol 93:279–289 J 88:1312–1318 38. Sangare O, Bastos AD, Venter EH, Vosloo W (2004) A first molecular 58. Prins HHT (1996) Ecology and behaviour of the African buffalo. Chapman & epidemiological study of SAT-2 type foot-and-mouth disease viruses in West Hall, London Africa. Epidemiol Infect 132:525–532 59. World Organisation for Animal Health: Handistatus II - OIE Animal Health 39. Bronsvoort BM, Radford AD, Tanya VN, Nfon C, Kitching RP, Morgan KL Information Database. http://web.oie.int/hs2/report.asp?lang=en. Accessed 8 (2004) Molecular epidemiology of foot-and-mouth disease viruses in the June 2015. Adamawa province of Cameroon. J Clin Microbiol 42:2186–2196 60. World Organisation for Animal Health: WAHID Interface - OIE World Animal 40. Ehizibolo DO, Perez AM, Carrillo C, Pauszek S, AlKhamis M, Ajogi I, Umoh JU, Health Information Database. http://www.oie.int/wahis_2/public/wahid.php/ Kazeem HM, Ehizibolo PO, Fabian A, Berninger M, Moran K, Rodriguez LL, Wahidhome/Home. Accessed 8 June 2015. Metwally SA (2014) Epidemiological analysis, serological prevalence and 61. Melletti M, Penteriani V, Boitani L (2007) Habitat preferences of the secretive genotypic analysis of foot-and-mouth disease in Nigeria 2008–2009. forest buffalo (Syncerus caffer nanus) in Central Africa. J Zool 271:178–186 Transbound Emerg Dis 61:500–510 62. Malhi Y, Adu-Bredu S, Asare RA, Lewis SL, Mayaux P (2013) African rainforests: 41. Fasina FO, Connell DR, Talabi OA, Lazarus DD, Adeleke GA, Olusanya TP, past, present and future. Philos Trans R Soc Lond B Biol Sci 368:20120312 Hernandez JA (2013) Foot-and-mouth disease virus strains and examination 63. Bengis RG, Kock RA, Fischer J (2002) Infectious animal diseases: the wildlife/ of exposure factors associated with seropositivity of cattle herds in Nigeria livestock interface. Rev Sci Tech 21:53–65 during 2007–2009. Prev Vet Med 109:334–342 64. Kock R, Kock M, de Garine-Wichatitsky M, Chardonnet P, Caron A (2014) 42. Gorna K, Houndje E, Romey A, Relmy A, Blaise-Boisseau S, Kpodekon M, Livestock and buffalo (Syncerus caffer) interfaces in Africa: ecology of Saegerman C, Moutou F, Zientara S, Bakkali Kassimi L (2014) First isolation disease transmission and implications for conservation and development. In: and molecular characterization of foot-and-mouth disease virus in Benin. Melletti M, Burton J (eds) Ecology, Evolution and Behaviour of Wild Cattle Vet Microbiol 171:175–181 Implication for Conservation. Cambridge University Press, Cambridge, UK, p 608 43. Ludi A, Ahmed Z, Pomeroy LW, Pauszek SJ, Smoliga GR, Moritz M, Dickmu S, 65. Ferguson KJ, Cleaveland S, Haydon DT, Caron A, Kock RA, Lembo T, Abdoulkadiri S, Arzt J, Garabed R, Rodriguez LL: Serotype diversity of Foot- Hopcraft JG, Chardonnet B, Nyariki T, Keyyu J, Paton DJ, Kivaria FM (2013) and-Mouth-Disease virus in livestock without history of vaccination in the Evaluating the potential for the environmentally sustainable control of foot far North Region of Cameroon. Transbound Emerg Dis, (in press) and mouth disease in Sub-Saharan Africa. Ecohealth 10:314–322 44. Ferris NP, Donaldson AI (1992) The World Reference Laboratory for Foot 66. Thomson GR (1995) Overview of foot and mouth disease in southern Africa. and Mouth Disease: a review of thirty-three years of activity (1958–1991). Rev Sci Tech 14:503–520 Rev Sci Tech 11:657–684 67. Ole-Mairon JO (2003) The Maasai ethnodiagnostic skill of livestock diseases: – 45. Roeder PL, Knowles NJ (2008) Foot-and-mouth disease virus type C a lead to traditional bioprospecting. J Ethnopharmacol 84:79 83 situation: the first target for eradication? In: FAO (eds) Open Session of the 68. IUCN: Syncerus caffer. In: IUCN Red List of Threatened Species Version European Commission for the Control of Foot-and-Mouth Disease Standing 2012.1. 2012. http://iucnredlist.org Accessed 25 February 2014. Technical Committee: 14–17 October 2008; Erice, Italy 69. IUCN UNEP-WCMC, IUCN, UNEP-WCMC (2014) The World Database on 46. Hedger RS, Forman AJ, Woodford MH (1973) Foot-and-mouth disease in Protected Areas (WDPA) [On-line]. UNEP-WCMC, Cambridge, UK East African buffalo. Bull Epizoot Dis Afr 21:90–99 47. Bastos AD, Anderson EC, Bengis RG, Keet DF, Winterbach HK, Thomson GR (2003) Molecular epidemiology of SAT3-type foot-and-mouth disease. Virus Genes 27:283–290 48. Dhikusooka MT, Tjornehoj K, Ayebazibwe C, Namatovu A, Ruhweza S, Siegismund HR, Wekesa SN, Normann P, Belsham GJ (2015) Foot-and- mouth disease virus serotype SAT 3 in long-horned Ankole calf, Uganda. Emerg Infect Dis 21:111–114 49. Hedger RS, Condy JB, Golding SM (1972) Infection of some species of African wildlife with foot-and-mouth disease virus. J Comp Pathol 82:455–461 50. Woodbury EL (1995) A review of the possible mechanisms for the persistence of foot-and-mouth disease virus. Epidemiol Infect 114:1–13 51. Hall MD, Knowles NJ, Wadsworth J, Rambaut A, Woolhouse ME (2013) Reconstructing geographical movements and host species transitions of foot-and-mouth disease virus serotype SAT 2. MBio 4:e00591–00513 52. Vosloo W, Bastos AD, Sangare O, Hargreaves SK, Thomson GR (2002) Review of the status and control of foot and mouth disease in sub-Saharan Africa. Rev Sci Tech 21:437–449 53. Vosloo W, Bastos ADS, Sahle M, Sangare O, Dwarka RM, Osofsky SA, Cleaveland S, Karesh WB, Kock MD, Nyhus PJ, Starr L, Yang A (2005) Virus topotype and the role of wildlife in foot and mouth disease in Africa. In: Conservation and development interventions at the wildlife/livestock interface: implications for wildlife, livestock and human health. IUCN, Gland, Submit your next manuscript to BioMed Central Switzerland and Cambridge, UK, pp 67–80 and take full advantage of: 54. Molla B, Ayelet G, Asfaw Y, Jibril Y, Ganga G, Gelaye E (2010) Epidemiological study on foot-and-mouth disease in cattle: seroprevalence • Convenient online submission and risk factor assessment in South Omo zone, south-western Ethiopia. Transbound Emerg Dis 57:340–347 • Thorough peer review 55. Miguel E, Grosbois V, Caron A, Boulinier T, Fritz H, Cornelis D, Foggin C, • No space constraints or color figure charges Makaya PV, Tshabalala PT, De Garine-Wichatitsky M (2013) Contacts and foot and mouth disease transmission from wild to domestic bovines in Africa. • Immediate publication on acceptance Ecosphere 4:51 • Inclusion in PubMed, CAS, Scopus and Google Scholar 56. Anderson EC, Foggin CM, Atkinson M, Sorensen KJ, Madekurozwa RL, • Research which is freely available for redistribution Nqindi J (1993) The role of wild animals, other than buffalo, in the current epidemiology of foot-and-mouth disease in Zimbabwe. Epidemiol Infect 111:559–563 Submit your manuscript at www.biomedcentral.com/submit